Kriging Great Lakes

Sediment Contamination
Values Cookbook for
ArcGIS 10.x

By

Michael Dennis, K. Wayne Forsythe, Cameron Hare, and

Adrian Gawedzki

2004-2012

To reference this document:

Dennis, M., K.W. Forsythe, C. Hare, and A. Gawedzki. 2012. Kriging Great Lakes Sediment
Contamination Values Cookbook for ArcGIS 10.x. Toronto, Ontario: Department of
Geography, Ryerson University. 51 pp.
Table of Contents

Table of Contents ............................................................................................................................ 1

Preparing the Data........................................................................................................................... 2
Running the Interpolation ............................................................................................................... 6
Variogram Model Selection ........................................................................................................ 7
Adjusting the symbology of the Kriging layer ......................................................................... 15
Creating the Vector Shapefiles ..................................................................................................... 24
Preparing the Layout ..................................................................................................................... 32
Adding the Legend .................................................................................................................... 41
Advanced Layout ...................................................................................................................... 47
Log Transformations ..................................................................................................................... 48

1
Preparing the Data

In previous versions of ArcMap, there existed an option to specify a no data value. Since all of
the missing values found in the point data (ie. 1998_Ont_pts.shp) are labeled as -1, this option
allowed for the exclusion of such points from the interpolation. However, in ArcMap 10, this
option no longer exists. If a kriging model is run, it will include all the values of a variable,
counting the missing data as -1, and will produce skewed results. Therefore, this data needs to
be modified, and the missing values must be removed from the dataset prior to being imported
into ArcMap.

Open Microsoft Excel, and click File Open. Find the source file where the point data is
located. At the bottom right of the Open window, ensure that dBase Files (*.dbf) is selected.
Double click the .dbf file of the point data to open it.

To identify all of the cells that have missing values of -1, hit Ctrl+F on the keyboard. This will
open the Find and Replace window. Click the Replace tab at the top. Type -1 into the Find
what: box. Leave the Replace with: box empty so that the cells that have a value of -1 will
become blank. Click Find All, and this will show all of the cells that have been identified. Click
Replace All, and all of the cells that previously had a value of -1 will now be blank. Click
Close.

2
Click File Save As, and save the file as a .xlsx to the desired destination with a suitable name
(such as 1998_Ont_pts_edit). Close Excel.

Open ArcMap. Click the Add Data button and double click the recently altered .xlsx file, and
double click again on the worksheet in which the data is stored (usually the file on top).

Since this is a spreadsheet file with no spatial dimension, it is necessary to create a new shapefile
based on the data within. To do this, click on the ArcToolbox icon, then expand Data
Management Tools Projections and Transformations, and double click Convert Coordinate
Notation.

3
Under Input Table, select the modified .xlsx spreadsheet. Under Output Feature Class, specify
the destination, and name the file appropriately. Under X Field (Longitude), select LON, and
under Y Field (Latitude), select LAT. Since the coordinate formats will not be used, Input
Coordinate Format and Output Coordinate Format can be left as default (DD). Under ID, select
STN_, as this will be the field used to later join the spatial data to the .xlsx. Under Spatial
Reference, select the appropriate projection (in this case, NAD_1983_UTM_Zone_17N).

Click OK.

A shapefile will be produced from the latitude and longitude coordinates found in the .xlsx file.
However, upon opening the attribute table for the shapefile, it can be seen that there is no
contaminant data. This can be repaired by right clicking the layer in the Table of Contents,
hovering over Joins and Relates and clicking Join The .xlsx table should be selected by
default in 2. Under 1., select the field that the join will be based on (STN_ in this example).
This should automatically select the same field in 3. Ensure Keep all records is selected.
Before clicking OK, the window should look like the following:

4
Click OK. Open the attribute table of the shapefile to ensure that the join worked. It should
now display all of the contaminants, and show <Null> where there are missing values. The
shapefile is now ready to be used for Kriging analysis.

5
Running the Interpolation

In ArcMap:

Add the following layers to the existing layers:

Shoreline vector (ie. realdeal_utm17.shp; Erie-UTM17.shp)

Clip layer (ie. ontario_clip.shp)

The point data contains the data used to perform kriging.

The shoreline vector layer symbology should be Hollow

Click the symbol once in the Table of Contents and click on the Hollow option.

The clip layer does not need to be visible as it is not used in the final product.

Select Customize Toolbars and ensure that the Geostatistical Analyst extension is active.
Click Launch Geostatistical Wizard to open the tool.

6
Variogram Model Selection

With Kriging/CoKriging highlighted under Methods, ensure that the Source Dataset is the
point data (ie.1998_Ont_pts), and the Data Field is the contaminant that you wish to
interpolate (mercury in this case).

Click Next.

7
Leave everything on the next screen as the defaults:

Click Next.

8
On the next screen, Geostatistical Wizard - Kriging Step 3 of 5 - Semivariogram/Covariance
Modeling:

Under Model #1, ensure that Spherical is selected as the Type.

Adjust the Major Range to 100,000 by clicking the calculator icon to the right of
the value (which will change to a pencil icon), then typing in the number.
Under Anisotropy, select True.
Adjust the Minor Range to 50,000 in the same manner as the Major Range.
Adjust the Direction box to 90 for Lake Ontario, or 70 for Lake Erie.
NOTE: These values are determined through experimentation and the distribution
characteristics of your points.

These attributes remain the same throughout model testing, which takes place next.

Click Next.

9
In Geostatistical Wizard - Kriging Step 4 of 5 - Searching Neighborhood, ensure that:

Maximum Neighbours is 5
Minimum Neighbours is 1
Sector Type is 4 Sectors with 45 degree offset
NOTE: These values are determined through experimentation and the distribution of
your points.

Click Next.

10
The following step performs the Cross Validation and calculates the prediction error statistics
Mean, Root-Mean-Square (RMS), Mean Standardized, Standardized Root-Mean-Square
Prediction Error (SRMSPE), and Average Standard Error (ASE).

Record these values in a table such as Error Stats Table.xls.

The |RMS-ASE| is not calculated by ArcGIS, but can be set up to be calculated automatically as
it has in the given table. This value is used for comparing the accuracy of the models tested.

Click Back until you arrive at the Geostatistical Wizard - Kriging Step 3 of 5 -
Semivariogram/Covariance Modeling window, and change the model to Exponential. You
will notice the blue lines on the Semivariogram graph move slightly. The Major and Minor
Ranges and the Direction should all remain at the values that were entered previously.

11
Click Next until you return to Geostatistical Wizard -Kriging Step 5 of 5 - Cross Validation
and record the error stats under the Exponential section of the table (the values in Step 4 should
remain unchanged).

Repeat these steps with the Gaussian model.

With the error stats for the three models entered into the Excel spreadsheet, the errors can be
compared between models, keeping in mind that:
The mean should be as close to 0 as possible,
The RMS and ASE should both be not more than 20,
The RMS and ASE should be similar, as calculated in the |RMS-ASE| column,
And the SRMSPE should be as close to 1 as possible.

Select the model that provides the most accurate interpolation (ie. the best error stats) and record
this in a new column in the table or simply highlight the model that has been selected within the
table.

12
Return to Step 3 of the Geostatistical Wizard, and click the model that you have just selected.
Click Next until you return to Step 5 and click Finish. In the Output Layer Information
window, ensure that all parameters are, in fact, correct.

Click OK. The prediction surface is now generated.

13
In the Table of Contents, move this new layer below the shoreline vector layer. Keep in mind
that this layer is temporary. Right click the layer and go to Properties. Under the Extent tab,
Set the extent to: the rectangular extent of realdeal-utm17. This will ensure that the Kriging
layer covers the extent of the shoreline vector. Click OK.

14
Adjusting the symbology of the Kriging layer

Double click on the Layer Name of the Kriging layer to open the properties window. Click the
Symbology tab, and then click on Filled Contours.

Check the Presentation Quality box, and then click on Classify to open the Classification
window.

15
Change the Classification Method to Manual, and adjust the number of classes (Classes) to
however many are necessary. This will be determined by the minimum and maximum values
and the intervals that are used. In this example of mercury, the intervals will be based on the
values of the TEL (0.174) and the PEL (0.486). There will be 3 equal intervals below the TEL, 3
equal intervals between the TEL and the PEL, and 3 intervals (based on the ranges between the
TEL and the PEL) above the PEL. Therefore there will be 9 custom classes, corresponding to
those found in Colours.xls. Adjust the breaks (on the right side) to incorporate the necessary
intervals.

16
Click OK in the Classification window.

In the Layer Properties window, Click and check off Contours on the left side of the window.
Both Contours and Filled Contours should be checked off.

As with the Filled Contours, check the Presentation Quality box.

17
Click on the Classify button to open the Classification window.

18
If the range of values encompasses both the TEL and the PEL values, change the number of
classes to 3. If the range of values includes only one of the TEL and PEL, change the number of
classes to two. In the case of mercury in Lake Ontario, both TEL and PEL are included.

Change the Method the manual, and change the Breaks values to equal the TEL and PEL
values:

19
Click OK, and return to the Layer Properties window.

20
Adjust the colours of these two lines. Double click the line representing the TEL (0.174, in this
case), and change its colour to green (Medium Apple, to be precise). Do the same with the PEL
(0.486) and make it red (Mars Red).

21
Adjust the colours for the filled contours as well. Click Filled Contours on the left menu. Based
on the RGB colour settings found in Colours.xls, adjust each of the classes accordingly by
double-clicking on the colour symbol. Under Fill Colour in the Symbol Selector window, select
More Colour. In the new window, Colour Selector, change to RGB (in the top right corner),
which will allow for the accurate input of the desired RGB values.

22
Alternatively, if a greyscale map is desired, while in the Colour Selector window, click the right-
facing arrow in the top-right corner and click Advanced Selection. Click the drop down menu
at the top, and select Gray. The desired grayscale value can then be entered for the class (also
found in Colours.xls).

23
Click OK in the Layer Properties window. The ArcMap window should now look as follows
(using the colour scheme and ranges in Colours.xlsx):

Creating the Vector Shapefiles

Since all predictions made outside of the lake are unnecessary, the vectors must be created from
the Kriging layer, then cropped (which is where the ontario_clip.shp layer comes into play).

First, right click on the Kriging layer, and select Data Export to Vector and the GA
Layer to Contour window will pop up.

24
Change the Contour Type to Filled Contour, and ensure that the new shapefile will be exported
to the proper folder and with the right file name (current convention is to name the file
HG_vector). Change Contour Quality to presentation, and click OK.

After the vector layer is added to the display, right click the layer, go to Properties Symbology
and click Features on the left side. Click Apply, then OK, and the ArcMap window should
look as follows:

Next, you want to get rid of all data outside of the lake. You can accomplish this by using Clip
in the ArcToolbox:

Select the ArcToolbox Analysis Tools Extract Clip

25
The input layer to clip is the one that was just created, HG_vector; the clip features is
ontario_clip. Write this file to the same folder as HG_vector, calling it HG_vector_clip, for
example. Leave the XY Tolerance empty, which is the default.

Click OK

Remove the HG_vector.shp layer. The map should now look as follows:

26
To create the isolines, right click on the Kriging Layer Name, select Data Export to
Vector, but this time, leave the Contour Type as Contour (as opposed to Filled Contour
previously).

Change the filename to HG_Isolines.shp (as has been used in all analyses thus far), ensuring that
the file is being written to the proper folder. Change Contour Quality to Presentation. Click
OK.

Again, the data outside of the lake must be deleted. Select the clipping tool. The input feature is
now HG_Isolines, and the clip feature remains ontario_clip. Change the filename to
HG_Isolines_clip.shp, ensuring it is written to the right folder.

Click Finish and remove the HG_Isolines layer from the Table of Contents.

The map should be as follows:

27
The Kriging layer can now be unchecked in the Table of Contents, making it invisible. The file
of the shoreline vectors should be moved to the top of the list in the Table of Contents.

The symbology of these two new layers must now be adjusted.

Double click on the HG_Isolines_clip layer to open Layer Properties. Go to the Symbology tab.

Click Categories on the left side of the window. Change the Value field from Classes to
Value and click the Add All Values button at the bottom of the window. In this case, only the
PEL isoline appears, because all TEL values lie outside of the lake. Double click on the line, and
change its colour to red (Mars Red) (or Medium Apple [green] if it is the TEL isoline). Uncheck
the <all other values> box.

Click in the Label field of the value, and change it to 0.486 ug/g (PEL) Isoline. These values
will vary, of course, depending on the contaminant, and whether or not the value is the TEL or
the PEL.

28
Click OK

Open the Layer Properties of the HG_vector_clip layer, and go to the Symbology tab.

Click Quantities on the left side of the window, and in the Fields box, change the value field to
Classes. If there are classes that are grouped together when this is done (a likely scenario),
simply increase the number of classes in the Classification box within the Symbology tab.

29
The classes must now be labeled.

To determine the value that each class represents, refer to the original Kriging Layer that was
created. Keep in mind that the numbering of classes in the Kriging layer begins with 0.
Therefore, class 3 in the HG_vector_clip represents 0.174 0.278 ug/g. Label each class 0.174
- <0.278 0.278 - <0.382 since it is always the lower break that is correct. Keep in mind that
the number of classes in the clipped file may not be the same number as in the Kriging layer
since some of the classes in the Kriging layer may be found exclusively outside of the lake, and
therefore not included in the clipped vector layer. Make sure to adjust the colours accordingly to
those found in Colours.xls.

30
Click on the Symbol column heading, and select Properties for All Symbols. Change the
Outline Width to 0.

Click OK.

Click OK in Layer Properties.

The map should now look as follows:

31
Click on the heading of the HG_vector_clip layer (where it says Classes right below the Layer
Name), click again, and change the heading to Prediction Intervals (ug/g)

**It is important to remember that when working with trace metals, measurements are in ug/g,
but with everything else (organochlorines and PAHs) measurements are in ng/g.

Youre finally ready to start preparing the actual layout.

Preparing the Layout

The points shapefile layer can be unchecked, as it will not be included on the final map. Click on
the layout button to the bottom left of the map.

Change the page layout to landscape by clicking File Page and Print Setup, then click
Landscape in the middle of the window. Click OK.

Right click on the data frame (the frame where the map is), select Properties, and go to the Size
and Position tab. The position and size coordinates should be as follows (ensure that the centre
anchor point is selected):

32
Click OK

On the Standard toolbar, change the zoom to 1:2,009,129.

Using the Pan tool (on the Tools toolbar), the map may need to be moved slightly.

So far, the layout should look like this:

33
Go back into the Data Frame Properties (right click on the data frame and select Properties), and
go to the Grids tab.

Click the New Grid button. Select Graticule, then Next.

34
In the Create a graticule window, select Labels only and change the Intervals to 1 degree for
both parallels and meridians.

35
Click Next.

In the Axes and labels window, change the Number of ticks per major division to 2.

36
Click Next.

Select Place a simple border at edge of graticule and click on the button below to adjust its
properties. Change the width of the line from 1.5 to 0.5.

Click Finish.

In Data Frame Properties, Click OK.

37
The title, legend, scalebar, and north arrow can now be added.

To add the title, go to Insert Title

Change the title to, for example, 1998 Lake Ontario Mercury Concentrations. Click OK.

Double click on the text to open the Properties window. In the Text tab, click the Change
Symbol button, and change the font to Times New Roman 18pt bold. Click OK.

Go to the Size and Position tab, ensure that the central Anchor Point is selected, and change the
X and Y Position coordinates to X: 5.5, Y: 6.63.

Click OK.

To insert the Neatline, go to Insert Neatline. In the Neatline window, select Place inside
margins and ensure that the Gap is 10 pts. Change the Background colour to Hollow
using the dropdown menu. Click OK.

38
To insert the scale bar, go to Insert Scale Bar. In the Scale Bar Selector window, click
Alternating Scale Bar 2.

Click the Properties button in the Scale Bar Selector window.

In the Units box in the Scale and Units tab, change the Division Units to Kilometers. In
Label change Kilometers to Kilometres (we are Canadian, after all). In the Scale box of
this window, check off the Show one division before zero box, and change the Number of
divisions to 2 and the Number of subdivisions to 4.

Go to the Format tab, and change the font to Times New Roman

Click OK, and OK again in the Scale Bar Selector window.

39
Click the scale bar and drag it to below the data frame.

With the scalebar selected, adjust the width until the major division (on the right) is 50 km by
clicking and dragging the box on the middle right or left.

Double click on the scale bar to open the Alternating Scale Bar Properties window. Go to the
size and position tab, and change the Position coordinates to X: 7.2, Y: 2.45, ensuring the central
Anchor Point is selected. Click OK.

To insert the north arrow, go to Insert North Arrow. Select ESRI North 1 and hit OK.

Double click on the north arrow to open the North Arrow Properties window. In the Size and
Position tab, change the position to X: 8.7, Y: 1.9, again ensuring the central Anchor Point is
selected. Click OK.

The layout should now look as follows:

40
Adding the Legend

To put the legend in, go to Insert Legend.

Remove all items in Legend Items by clicking the button.

In Map Layers select HG_vector_clip and hit the button to add it to Legend Items.
Do the same with realdeal-utm17 (the shoreline vector dont ask about the naming), and
HG_isolines_clip. Change the Set the number of columns in your legend to 3.

41
Click Next

In Legend Title font properties change the Size to 12, the font to Times New Roman, and
make it bold and underlined.

42
Click Next until you get to the final window of the Legend Wizard, and finally click Finish

Move the legend to below the dataframe, and not on top of the scale bar or north arrow. The
legend should currently look like this:

Double-click the legend to open the Legend Properties window. Under the Items tab, make
sure each Legend Item has the Place in a new column box checked, and the Columns: for
each is set at only 1 (both at the right of the window). In Legend Items, double click on
HG_vector_clip, opening the Legend Item Selector window. Click the Properties button.

43
In the General tab, deselect Show Layer Name. Click the Heading Symbol button, and
change the font to Times New Roman 10 pt bold, and click OK. Click the Label Symbol
button, and change the font to Times New Roman 10 pt, and click OK. Click OK in the
Legend Item window, and OK in the Legend Item Selector window, returning to legend
properties.

Double click realdeal-utm17 in Legend Items. Using the same process, ensure that only Show
Labels is selected and change the Label Symbol to Times New Roman 10 pt.

Do the same for HG_Isolines_clip.

The legend should now look like this:

In the Table of Contents, click the Layer Name of the realdeal-UTM17 layer and change it to
No Data. This provides the No Data symbol for the legend a kind of round-about way of
doing things, but it works

Check to make sure that everything in the legend is correct prediction intervals, labels, etc. and
make any changes necessary.

Once the legend is correct, right click on it and click Convert to Graphics. Right click on it
again and click Ungroup.

The legend should now look like this:

44
You will want to use 4 rows of symbols (5 if there are a lot of classes). What you will need to do
to properly organize the legend is first, move the No Data and Isolines out of the way for the
time being by selecting and dragging them.

Next, in this case, you will want to move the fifth class, 0.8 -<1.0 to the top of the second
column. Next, move No Data under the 0.8 - <1.0 class, and the isoline(s) below this (it may be
necessary to have more than two columns, depending on the number of classes and isolines):

To ensure that all the items are properly aligned vertically, use the Y coordinates from items in
the left column and copy and paste these page coordinates to those in the other columns. For
example, to ensure alignment of the Isoline, in this case, double click on ,
opening the Properties window, highlight and copy the Y Position coordinate, ensuring that the
Anchor Point is centred vertically. Close down this Properties window (hit Cancel) and
double click the Isoline item, opening its property window. Highlight the Y Position coordinate
and paste the new one (from ). Hit OK. Repeat this for all remaining
items.

To ensure the proper spacing of columns, open the Properties window of the longest item in the
left column. Select right-central Anchor Point:

Add 0.2 inches to the X Position Coordinate

(ie. 3.281592 in becomes 3.481592). Copy
this coordinate, and hit Cancel in the
Properties window (you dont actually want to

45
apply this change to this legend item). Open the first legend item in the column to the right,
change the Anchor Point to the left-central point. Paste the coordinate that you copied from the
left column to the X Position coordinate of this item. Repeat this for all legend items in this
column. If there are more than two columns in the legend, these steps will need to be repeated.

Select all legend items, right click, and select Group to group them back into one element (this
ensures that no items will be accidentally moved relative to each other).

Double click on the legend, opening its Properties window. Change the Anchor Point to the top
left, and change the X Position coordinate to 2.122244 in (aligned with the left extent of the data
frame), and the Y Position coordinate to 2.5. Click OK.

At this point, you will probably want to save your work (since the layout is now complete) if you
havent done so already.

46
Go to File Save (or Save As)you know the rest

This will save it as an .mxd file, so your layout and attributes will all be saved.

Advanced Layout

These Position coordinates will change, depending on the size of the map that is being produced.
Generating these coordinates is not an easy task, and requires a lot of trial and error to achieve
the proper size and layout of everything.

The sizes that have been produced already are 11.5cm, 13cm, and full page (ie. for Powerpoint
presentations).

To change maps to the 11.5 cm or 13 cm sizes, the full page map must first be produced.
Change all of the Position Coordinates and sizes (Width and Height in the Properties
windows) to those included in FullPageXY.xls.

With full page maps made, the first step in making 11.5 cm or 13 cm maps is to change the page
layout to portrait (File Page and Print Setup). Delete the neatline. Then, select all of the
elements on the layout (title, dataframe, legend, north arrow, scale bar), right click and open the
Common Properties for All Elements window. In the Size and Position tab, change the Width
to i.) 4.2 for the 11.5 cm, or ii.) 4.75 for 13 cm.

Deselect all elements, and go into the Data Frame Properties window. Click on Grids, and click
Properties. Go to the Labels tab, and at the bottom, in the Label Orientation box, check off
Right and Left. Hit OK and OK.

With this done, change all Position coordinates and sizes to those in 11-5cmXY.xls and
13cmXY.xls.

It is important to keep in mind that when you reopen mxds, ArcMap often likes to slightly move
the layout positions of the scale bar and north arrow. Check to make sure that these are correct
before producing anything for publication.

To export maps as JPEGs or TIFFS, when in layout view in ArcMap, go to File Export.
Select the file type. Click the Options button and change parameters as necessary (ie. if
creating a JPEG, using a minimum quality of 200 dpi).

In some cases, you may need to manually alter the colours of the vector files (especially when
comparing data with different years or lakes). Some of these colour schemes have already been
generated, and can be found in Colours.xls. To manually change the colours, click on the
symbol that you want to change in the Table of Contents. This will open the Symbol Selector
window. Click the Fill Colour dropdown menu, and click More Colours at the bottom. Adjust
the RGB values accordingly.

47
For the Isolines, if producing B&W maps, change the PEL isoline to a solid black line by
clicking on the line in the Table of Contents, opening the Symbol Selector window. Change the
colour to black, and ensure that the Width is 1. For the TEL, change the symbol type to Dashed
4:4 (you will need to scroll down through all of the symbols in the Symbol Selector window to
find it). Change the colour to black and width to one.

Log Transformations (read all sections before performing)

Log transformations are necessary when the value of the ASE is greater than 20. There is more to
it than this but the previous statement can be used as a general rule. When the ASE is greater
than 20, actual prediction values stray quite far from the original input values. To do the log
transformations, do not use the log function within ArcMap. To do them, open up the .dbf file
of the point data in SPSS.

*Use the original file, not the file with the missing values removed made in Preparing the Data.

In the Data View of SPSS, click on Transform Compute Variable. In the Target Variable
box, enter a new variable name (ie. HgLog). In the functions, find LG10 (found under
Function Group: Arithmetic). Double click, and the function LG10(?) will show up in the
Numeric Expression box, with the question mark highlighted. Replace the ? with the
variable that you want to perform the transformation on (double click the variable from the list in
the left of the Compute Variable window).

Click on the If button, and in the new window check off Include if case satisfies
condition:. Double click the variable that you are performing the transformation on, entering it
into the expression box. Type in <> -1. *Ensure that there is a space between <> and -1*.
Click Continue. This will stop SPSS from trying to log a negative value which, of course, is
impossible.

In the Compute Variable window, click OK.

48
All of the new values will be input to the new variable that you defined, HgLog. However,
there will be some blank spaces where there was a -1 in the original data. Fill in these spaces
with -1s so that when you perform the kriging analyses on the data, these points will be omitted
from the analyses. However, ensure that the points that had original data values of 0 (and not
missing values) are returned to 0, and not counted as missing values (-1).

Alsowhen doing log-transforms, you need to do the following....

Missing values must be assigned (-1 or -9999) where the original data points have a value of 0.
Reason: you can't log 0, so this will allow them to be omitted from the transformation. In some
cases where a lot of 0s are in the files.there may not be enough points left to do a kriging
analysis on the log-transformed data.

Note: -1 is the number for missing values in the contemporary Erie and Ontario datasets.-9999
is used in all other files.

49
Click on File change the type to .dbf, and overwrite the original dbf. It may be necessary to
repeat the steps in the Preparing the Data section found at the beginning, since loading the
original .dbf into ArcMap may result in the missing values being reverted into zeros, rather than
blank cells. After performing the log transformation on the selected contaminant(s), repeat the
steps in Preparing the Data before performing the kriging analysis.

When running the kriging analysis, be sure to remember which log values represent which non-
log values so that the legends can be altered as necessary, i.e. log numbers would be represented
by their non-log equivalents in the map legend.

50